ac3enc: use generic fixed-point mdct
This makes the AC3 encoder use the shared fixed-point MDCT ratherthan its own implementation. The checksum changes are due todifferent rounding in the MDCT.
Signed-off-by: Mans Rullgard <email@example.com>
Add apply_window_int16() to DSPContext with x86-optimized versions and use itin the ac3_fixed encoder.
ac3enc: do not right-shift fixed-point coefficients in the final MDCT stage.
This increases the accuracy of coefficients, leading to improved quality.Rescaling of the coefficients to full 25-bit accuracy is done rather thanoffsetting the exponent values. This requires coefficient scaling to be done...
ac3enc: change default floor code to 7.
This is to match the value in every (E-)AC-3 file from commercial sources.It has a negligible effect on audio quality.
ac3enc: Change EXP_DIFF_THRESHOLD to 500.
This patch changes the exponent difference threshold in the exponentstrategy decision function of the AC-3 encoder. I tested lowering inincrements of 100. From 1000 down to 500 generally increased in qualitywith each step, but 400 was generally much worse....
Update mpegts test reference
The output was changed by a7827a17c6b3388322350456d445c94b3a82cd25.
mpegtsenc: set reserved bits to 1 in PCR field
The reserved bits between PCR base and extension fields must beset to 1.
Add regression test for stereo s16le in voc.
Change the default dB-per-bit code from 2 to 3.This gives slightly better quality in PEAQ tests.Code 3 gives a dBpb value of 2816 = -132dB (128 psd units = -6dB), whichcorresponds to 22 bits. Since the exponents have an offset applied, the16-bit source looks like 24-bit source to the bit allocation routine....
Change FIX15 back to clipping to -32767..32767.This avoids a 16-bit overflow in mdct512() due to a -32768 value in costab.References updated for acodec-ac3, lavf-rm, and seek-ac3_rm tests.Thanks to Måns Rullgård for finding the bug.
Originally committed as revision 26071 to svn://svn.ffmpeg.org/ffmpeg/trunk
Update the test references for lavf-rm and seek-ac3_rm.The references changed due to r25956.
Originally committed as revision 26004 to svn://svn.ffmpeg.org/ffmpeg/trunk
Avoid negative SCR in mpeg ps muxer.Fixes a scr issue reported with dvdauthor ([FFmpeg-user] FFMPEG encoded MPEG-2 video causes error in DVDAuthor)
Originally committed as revision 25512 to svn://svn.ffmpeg.org/ffmpeg/trunk
Update gxf regression tests because of r25399
Originally committed as revision 25400 to svn://svn.ffmpeg.org/ffmpeg/trunk
In gxf muxer, fix flt entry offset, patch by Reuben Martin, reuben dot m at gmail dot com
Originally committed as revision 25395 to svn://svn.ffmpeg.org/ffmpeg/trunk
Init SCR in mpeg muxer based on first DTS.This fixes issues if the first DTS is far away from 0.
Originally committed as revision 25383 to svn://svn.ffmpeg.org/ffmpeg/trunk
Test decoding in fate-lavf-gif
Originally committed as revision 25012 to svn://svn.ffmpeg.org/ffmpeg/trunk
PNG image regression testing
Originally committed as revision 25011 to svn://svn.ffmpeg.org/ffmpeg/trunk
gxfenc: Fix ES name in the UMF media description, by using strlen instead of sizeof
Patch by Thierry Foucu, tfoucu at gmail
Originally committed as revision 24379 to svn://svn.ffmpeg.org/ffmpeg/trunk
In mxf muxer, fix last b frame temporal offset
Originally committed as revision 23967 to svn://svn.ffmpeg.org/ffmpeg/trunk
matroskaenc: Don't write a second seekhead for the clusters; mkvalidate agreeswith me that it's unnecessary.
Originally committed as revision 23478 to svn://svn.ffmpeg.org/ffmpeg/trunk
matroskaenc: Mux clusters better
Start them on keyframes when reasonable, and delay writing audio packetsto help ensure that there's audio samples available for the first frame inclusters.
Patch by James Zern <jzern at google>
Originally committed as revision 23473 to svn://svn.ffmpeg.org/ffmpeg/trunk
matroskaenc: Write codec time base as default duration for video tracks.
This isn't exactly semantically equivalent, but the field has already beenlong abused to mean this, and writing it helps in determining a decent cfrtime base when transcoding from a mkv where the video codec stores none (VP8)....
Update regression tests after removing track timecode scale from mkvenc
Originally committed as revision 23248 to svn://svn.ffmpeg.org/ffmpeg/trunk
In ogg muxer, pack multiple frames into one page, much lower overhead
Originally committed as revision 23231 to svn://svn.ffmpeg.org/ffmpeg/trunk
Change default for bidir_refine to 1.
Originally committed as revision 22778 to svn://svn.ffmpeg.org/ffmpeg/trunk
10l, update ref value for ffm since default flags changed and are stored in the file
Originally committed as revision 22673 to svn://svn.ffmpeg.org/ffmpeg/trunk
use mpeg2 quantization bias for mjpeg.this seems to improve RD performance.
Originally committed as revision 22550 to svn://svn.ffmpeg.org/ffmpeg/trunk
mpegts vbr muxing, activated when muxing rate is not supplied by theuser.
Originally committed as revision 22352 to svn://svn.ffmpeg.org/ffmpeg/trunk
Start continuity counter at 0 for pmt as well
Originally committed as revision 22350 to svn://svn.ffmpeg.org/ffmpeg/trunk
Start continuity counter at 0 for streams
Originally committed as revision 22344 to svn://svn.ffmpeg.org/ffmpeg/trunk
10l typo, fix ts total bit rate computation
Originally committed as revision 22343 to svn://svn.ffmpeg.org/ffmpeg/trunk
regtest: place image sequence outputs in separate directories
Originally committed as revision 22282 to svn://svn.ffmpeg.org/ffmpeg/trunk
Place regtest-pixfmt output files in separate dir
This test generates many output files, and keeping them separateis convenient.
Originally committed as revision 22157 to svn://svn.ffmpeg.org/ffmpeg/trunk
Place regression test output files in subdirs per family
Originally committed as revision 22155 to svn://svn.ffmpeg.org/ffmpeg/trunk
Correctly increment continuity_counter in PCR packets.
Patch by Yann Coupin, yann.coupin+ffmpeg gmail
Originally committed as revision 22020 to svn://svn.ffmpeg.org/ffmpeg/trunk
Correct opts calulation in ffmpeg.c.This correct the stop point for demuxing with -vcodec copy and -t as well aspacket interleaving. (we already diddrop packets but kept demuxing themfor too long due to opts being wrong)the change to ffm is due to 2 packets with timestamp 0 being stored...
Check pkt.pts against the recording time.This fixes at least ogg encoding with -t where the file was slightly too long.
Originally committed as revision 21598 to svn://svn.ffmpeg.org/ffmpeg/trunk
Update regression tests after r21372
Originally committed as revision 21374 to svn://svn.ffmpeg.org/ffmpeg/trunk
regtest: split reference files allowing tests to run individually
With this change, the output is checked immediately after each testhas run. This means commands like "make regtest-mpeg2" can now beused to run a single test and get meaningful results....